Fork me on GitHub

Nanobody epitopes on SARS-CoV-2 spike protein

Integrative modeling of nanobody binding modes to the SARS-CoV-2 Spike protein PubMed logo

tickVerified to work with the latest stable IMP release (2.21.0). The files are also available at GitHub.
Additional software needed to use these files: IMP numpy pandas matplotlib biopython networkx scikit-learn hdbscan MODELLER install instructions

Anaconda logo To install the software needed to reproduce this system with the Anaconda Python command line tool (conda), run the following commands:

conda config --add channels salilab
conda install imp numpy pandas matplotlib biopython networkx scikit-learn hdbscan modeller

UCSF logo To set up the environment on the UCSF Wynton cluster to run this system, run:

module load Sali
module load imp python3/numpy python3/pandas python3/matplotlib python3/biopython python3/networkx python3/scikit python3/hdbscan modeller
Tags chemical crosslinks cryo-EM escape mutations nanobodies PMI shape-complementarity


This repository contains comparative models of 21 nanobodies and integrative models of their epitopes on the receptor-binding (RBD) and ectodomains of the SARS-CoV-2 spike protein. Epitopes were modelled using chemical crosslinks and escape mutagenesis data. Both receptor (spike protein) and nanobodies were represented as completely rigid subunits. This work develops a computationally efficient shape complementarity restraint focused around the escape mutant residues, distance restraints between nanobody CDR loops and viral escape residues and a modified interface-metric (inspired by the fcc metric) for clustering alternate models from structural sampling.

Both the receptor and nanobodies in this work have been coarse-grained at a single residue per coarse-grained bead, and subsequently subjected to rigid-rigid docking. Thus the exact orientation of the nanobody on the spike surface maybe noisy and need future refinements. The focus of this exercise is thus to predict a comprehensive epitope on the spike surface that is maximally consistent with input crosslink and escape data.

Nanobody names in this repository are simplified versions of those used in the paper.

Nanobody name in this repository Nanobody name in paper x
rbd-x S1-RBD-x 9, 15, 16, 21, 22, 23, 24, 29, 35, 40
s1-x S1-x 1, 6, 23, 36, 37, 46, 48, 49, 62
s2-x S2-x 10, 40

List of files and directories:


This is an integrative epitope modeling module written using IMP and PMI. It contains:


MODELLER scripts and top scoring comparative models of all 21 nanobodies both before and after loop refinement. All 21 nanobodies are modelled from the human Vsig4 targeting nanobody Nb119.


Contains nanobody structures generated by AlphaFold-2. The ColabFold notebook framework was used to run AlphaFold, specifically the AlphaFold2_batch Colab notebook with default settings, and structure relaxation post prediction. For each nanobody, the top ranked nanobody according to average plddt was selected for epitope modeling.


Contains scripts that use the nblib module to structurally sample, cluster and calculate restraint satisfaction for nanobody binding modes on the spike protein. Sub-folders:


The overall content is similar as above. Main differences are -


Author(s): Tanmoy Sanyal

Date: December 4, 2021

License: CC BY-SA 4.0 This work is licensed under the Creative Commons Attribution-ShareAlike 4.0 International License.

Last known good IMP version: build info build info

Testable: Yes.

Parallelizable: Yes